Abstract: Improving the forecasting skills of the meso-scale convective system (MCS) is one of the key scientific problems in the field of numerical weather prediction. The occurrence and development of severe convective weather are affected by multiple factors such as atmospheric thermodynamic and kinetic conditions, topography, and air pollution conditions. Due largely to the uncertainty in models and the inevitable errors of initial values, large uncertainties still exist for the accurate prediction of severe weather produced by the MCS. Therefore, to effectively improve the accuracy of severe convective weather forecasts in China, conducting targeted observation experiments in key areas of interest, which typically helps to reduce the uncertainty level of the model's initial meteorological field, may be one of the effective ways forward. It follows that the initiation and formation mechanisms of MCS can be revealed, and the forecast skills of severe convection will be improved. In this paper, we first propose the technology roadmap of targeted observation as follows. Based on the modern meteorological observational network over the Jing-Jin-Ji area, the typical MCS forecast sensitive regions are identified by using the Ensemble Transform Kalman filter, combined with the atmospheric sounding systems mounted on the mobile vehicle, where targeted observation experiments will be conducted. As such, convective initiation mechanisms are elucidated, and the novel methods expected to improve the forecast skill for MCS are explored. Secondly, in response to the challenge of "finding needles in a haystack" in the vertical observation of the lower atmosphere for short-term forecasting and warning of severe convective weather in large cities, the potential application value of a dynamic triangular observation mesonet in the study of triggering and development mechanisms of severe convective weather was explored through the construction of a Pyramid-shaped LOwer Tropospheric Observational System (PLOTOS) that consists of five stations with simultaneous vertical observation capabilities. Finally, it is suggested that the initiation and formation mechanisms of severe convective weather be unraveled using PLOTOS, and the mesoscale targeted observation technology be developed, which is crucial to the improvement of weather observational networks in large cities and provide new ideas and methods for improving the forecast of severe convective weather processes.